The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing the gh (GitHub CLI) and git commands to manipulate repository state, manage branches, and interact with the GitHub API.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the github/gh-stack extension via the GitHub CLI. This process downloads and executes extension code from GitHub's official repository, which is a recognized and trusted source.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from GitHub (such as pull request metadata and branch status) which represents a potential attack surface for indirect instructions from external contributors.
Ingestion points: The agent is instructed to parse the JSON output of gh stack view --json, which contains pull request titles and metadata from the remote repository.
Boundary markers: The skill lacks explicit instructions or delimiter patterns to ensure the agent ignores potentially adversarial content embedded within pull request descriptions or titles.
Capability inventory: The skill provides capabilities to execute multiple shell commands (gh, git) and perform file system operations (git commits/adds) based on the processed metadata.
Sanitization: No specific sanitization or validation logic is defined for the data retrieved from the GitHub API before it influences agent logic.

gh-stack